[Bug] CPU usage does not match data shown in Web

ejpenney commented 11 months ago

The problem

As mentioned in #677, CPU information shown in the ASUS dashboard is different than data shown in Home Assistant.

Update interval is default (30s). Here's an hour chosen from yesterday:

Interestingly, immediately after this hour, the CPU usage went to near zero. It's hard to say for sure, but it's almost like, if the usage goes over ~50%, then HA reports it as 100%. It's either up or down, no in-between? But those are just my initial observations, looking at it for all of 5 minutes.

Here we see everything was fine, then I pulled up the Asus page in a browser and started looking at them side-by side:

I tried to get an hour of logbook data for you, but it appears the CPU sensor isn't supported. Hopefully these screenshots are useful? If not, I'm happy to add more data.

Original screenshots:

Your device model

RT-AC66U

Firmware type

Stock

Firmware version

3.0.0.4.380.70_0

Integration version with the issue

0.24.0

Method of the integration installation

HACS

What version and type of Home Assistant installation do you use

2023.11.2

Is there anything useful in the logs?

No response

Diagnostics information

No response

Additional information

No response

Vaskivskyi commented 11 months ago

It seems to be rather hard to track this issue. So, I'll be posting some random questions to try to guess where to look for the reason.

Do you have a single Asus device connected to HA?
Can you change the update time to e.g. 5 s for some time (also need to change cache time to 4 s or less. Will it be jumping up-down or stably showing such high values?

ejpenney commented 11 months ago

No problem, let me know any way I can help.

Yes, it's just the one single device connected to Home Assistant.
I changed update time to 5s and debug cache to 4s and have let it run a few days. Looks to me like a little of both?

Here's an hour from this morning:

Valdorama commented 10 months ago

Hello - just want to add that I've been seeing this issue too every now and then. The integration seems to report that the CPU is much higher than it actually is (as reported by the router web admin page). As an example, here you can see the integration suggesting the CPU has been consistently in the 40-50% range for over 12 hours: